On 5/2/24 12:34, Eduard Zingerman wrote:
On Wed, 2024-05-01 at 13:47 -0700, Kui-Feng Lee wrote:
The verifier has field information for specific special types, such as
kptr, rbtree root, and list head. These types are handled
differently. However, we did not previously examine the types of fields of
a struct type variable. Field information records were not generated for
the kptrs, rbtree roots, and linked_list heads that are not located at the
outermost struct type of a variable.
For example,
struct A {
struct task_struct __kptr * task;
};
struct B {
struct A mem_a;
}
struct B var_b;
It did not examine "struct A" so as not to generate field information for
the kptr in "struct A" for "var_b".
This patch enables BPF programs to define fields of these special types in
a struct type other than the direct type of a variable or in a struct type
that is the type of a field in the value type of a map.
Signed-off-by: Kui-Feng Lee <thinker.li@xxxxxxxxx>
---
I think that the main logic of this commit is fine.
A few nitpicks about code organization below.
kernel/bpf/btf.c | 118 +++++++++++++++++++++++++++++++++++++++--------
1 file changed, 98 insertions(+), 20 deletions(-)
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 4a78cd28fab0..2ceff77b7e71 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -3493,41 +3493,83 @@ static int btf_get_field_type(const char *name, u32 field_mask, u32 *seen_mask,
[...]
+static int btf_find_struct_field(const struct btf *btf,
+ const struct btf_type *t, u32 field_mask,
+ struct btf_field_info *info, int info_cnt);
+
+/* Find special fields in the struct type of a field.
+ *
+ * This function is used to find fields of special types that is not a
+ * global variable or a direct field of a struct type. It also handles the
+ * repetition if it is the element type of an array.
+ */
+static int btf_find_nested_struct(const struct btf *btf, const struct btf_type *t,
+ u32 off, u32 nelems,
+ u32 field_mask, struct btf_field_info *info,
+ int info_cnt)
+{
+ int ret, err, i;
+
+ ret = btf_find_struct_field(btf, t, field_mask, info, info_cnt);
btf_find_nested_struct() and btf_find_struct_field() are mutually recursive,
as far as I can see this is usually avoided in kernel source.
Would it be possible to make stack explicit or limit traversal depth here?
Sure!
The 'info_cnt' field won't work as it could be unchanged in
btf_find_struct_field() if 'idx == 0'
+
+ if (ret <= 0)
+ return ret;
+
+ /* Shift the offsets of the nested struct fields to the offsets
+ * related to the container.
+ */
+ for (i = 0; i < ret; i++)
+ info[i].off += off;
+
+ if (nelems > 1) {
+ err = btf_repeat_fields(info, 0, ret, nelems - 1, t->size);
+ if (err == 0)
+ ret *= nelems;
+ else
+ ret = err;
+ }
+
+ return ret;
+}
+
static int btf_find_struct_field(const struct btf *btf,
const struct btf_type *t, u32 field_mask,
struct btf_field_info *info, int info_cnt)
[...]
@@ -3644,6 +3707,21 @@ static int btf_find_datasec_var(const struct btf *btf, const struct btf_type *t,
field_type = btf_get_field_type(__btf_name_by_offset(btf, var_type->name_off),
field_mask, &seen_mask, &align, &sz);
Actions taken for members in btf_find_datasec_var() and
btf_find_struct_field() are almost identical, would it be possible to
add a refactoring commit this patch-set so that common logic is moved
to a separate function? It looks like this function would have to be
parameterized only by member size and offset.
Of course, it could be.
+ /* Look into variables of struct types */
+ if ((field_type == BPF_KPTR_REF || !field_type) &&
+ __btf_type_is_struct(var_type)) {
+ sz = var_type->size;
+ if (vsi->size != sz * nelems)
+ continue;
+ off = vsi->offset;
+ ret = btf_find_nested_struct(btf, var_type, off, nelems, field_mask,
+ &info[idx], info_cnt - idx);
+ if (ret < 0)
+ return ret;
+ idx += ret;
+ continue;
+ }
+
if (field_type == 0)
continue;
if (field_type < 0)
[...]
